Research Article Feasibility of RNA and DNA Extraction ...downloads.hindawi.com/journals/ogi/2013/576842.pdf · fresh or fresh frozen samples, as the purity and quantity of extracted

Hindawi Publishing CorporationObstetrics and Gynecology InternationalVolume 2013, Article ID 576842, 9 pageshttp://dx.doi.org/10.1155/2013/576842

Research ArticleFeasibility of RNA and DNA Extraction fromFresh Pipelle and Archival Endometrial Tissues forUse in Gene Expression and SNP Arrays

Heather D. Kissel,1 Thomas G. Paulson,1 Karen Liu,2 Xiaohong Li,1

Elizabeth Swisher,3 Rochelle Garcia,4 Carissa A. Sanchez,1 Brian J. Reid,1,5,6,7

Susan D. Reed,3,5 and Jennifer Anne Doherty5,8,9

1 Division of Human Biology, Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA2Division of Vaccine and Infectious Disease, Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA3Department of Obstetrics and Gynecology, University of Washington School of Medicine, Seattle, WA 98105, USA4Department of Pathology, University of Washington Medical School, Seattle, WA 98105, USA5Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA6Department of Medicine, University of Washington, Seattle, WA 98105, USA7Department of Genome Sciences, University of Washington, Seattle, WA 98105, USA8The Geisel School of Medicine at Dartmouth, Hanover, NH 03755, USA9The Norris Cotton Cancer Center, Lebanon, NH 03756, USA

Correspondence should be addressed to Jennifer Anne Doherty; [email protected]

Received 8 May 2013; Accepted 22 August 2013

Academic Editor: Eric Prossnitz

Copyright © 2013 Heather D. Kissel et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.

Identifying molecular markers of endometrial hyperplasia (neoplasia) progression is critical to cancer prevention. To assess RNAandDNAquantity and quality from routinely collected endometrial samples and evaluate the performance of RNA- andDNA-basedarrays across endometrial tissue types, we collected fresh frozen (FF) Pipelle, FF curettage, and formalin-fixed paraffin-embedded(FFPE) hysterectomy specimens (benign indications) from eight women. Additionally, neoplastic and uninvolved tissues from 24FFPE archival hysterectomy specimens with endometrial hyperplasias and carcinomas were assessed. RNA was extracted from 15of 16 FF and 51 of 51 FFPE samples, with yields >1.2𝜇g for 13/15 (87%) FF and 50/51 (98%) FFPE samples. Extracted RNAwas of highquality; all samples performed successfully on the Illumina whole-genome cDNA-mediated annealing, selection, extension, andligation (WG-DASL) array and performance did not vary by tissue type. While DNA quantity from FFPE samples was excellent,quality was not sufficient for successful performance on the Affymetrix SNPArray 6.0. In conclusion, FF Pipelle samples, which areminimally invasive, yielded excellent quantity and quality of RNA for gene expression arrays (similar to FF curettage) and shouldbe considered for use in genomic studies. FFPE-derived DNA should be evaluated on new rapidly evolving sequencing platforms.

1. IntroductionThough endometrial carcinoma is the most common gyne-cologic malignant neoplasm [1], diagnostic capabilities andmanagement of endometrial precancer (intraepithelial neo-plasia) lag far behind those of cervical carcinoma [2]. Aneoplastic continuum from simple, to complex, to atypicalhyperplasia, to endometrial carcinoma is suggested fromlongitudinal epidemiologic studies [3–5]. Identification of

molecular alterations present in various stages of endometrialneoplasia will provide the basis for early detection and thera-peutics [6]. Present diagnostic capabilities utilizing histologicevaluation for endometrial hyperplasia/neoplasia alone arelimited by poor diagnostic reproducibility [7] and relativelylow prognostic value. Risk of progression to carcinomaamong women with a diagnosis of endometrial hyperplasiawith atypia is not well understood, though exposure to

2 Obstetrics and Gynecology International

progestin therapy has been reported to be associated withan approximately 60% decreased risk of progression [4, 8].Future studies that attempt to elucidatemolecular biomarkersof endometrial hyperplasia progression risk will requirethe development of two methodologies: the ability to per-form array studies from extremely small fresh endometrialsamples, as well as high-fidelity large-scale affordable arrayinterrogation of archived FFPE samples. Potential challengesexist with both.

The growing field of genomic technologies has madelarge array-based studies of somatic mutations possible;whole exome sequencing and mutation detection have led toidentification of potential driver cancer genes in endometrialcancer [9, 10]. Tissue archives of longitudinal endometrialneoplasia (intraepithelial neoplasm) specimens represent apotentially valuable resource for genomic studies, but theyare typically comprised of formalin-fixed paraffin-embedded(FFPE) samples [4, 5]. High-quality nucleic acids necessaryfor gene expression profiling are most readily obtained fromfresh or fresh frozen samples, as the purity and quantityof extracted RNA and DNA from archival tissues can behighly variable [11]. However, the whole-genome cDNA-mediated annealing, selection, extension, and ligation (WG-DASL) Assay (Illumina, San Diego, CA, USA) [12] hasbeen implemented using partially degraded RNA extractedfrom FFPE tissues for breast carcinoma [13–15] and ovar-ian carcinoma [16], with varying degrees of success. It isessential to evaluate the quantity and quality of RNA andDNA obtained from endometrial FFPE samples and theirperformance on high-throughput arrays to determine thefeasibility of using these readily available tissue archives tofurther our understanding of longitudinal progression ofendometrial neoplasms and improve diagnostic accuracy andthereby, therapeutic choices.

Women diagnosed with endometrial neoplasia presentwith abnormal or heavy bleeding patterns and are usuallyfirst evaluated with office endometrial biopsies [17, 18], oftensampled atmultiple time points. Samples are obtained blindlyusing a disposable suction device placed into the uterus,which results in random sampling that is not oriented touterine site. The most common device is a Pipelle. The smalldiameter (3mm) and flexible plastic suction curette improvetolerability of this office procedure. Volumes of tissue samplesobtained are usually 1-2 cubic centimeters (cc) or less than1 cc in atrophic samples, and can be greater than 3 cc inproliferative samples, for example, endometrial hyperpla-sia/neoplasia. Moving forward, in the clinical scenario ofa woman presenting with abnormal bleeding, we envisionholding a small amount of fresh office endometrial biopsyspecimen for biomarker analysis while retaining the bulk ofthe specimen for FFPE clinical diagnostic analysis.

To address whether the quantity and quality of RNAobtained from office endometrial samples and archived FFPEendometrial tissues were suitable for downstream arrayanalyses, we compared the quality and quantity of RNAisolated from three endometrial specimen types from thesame individual (a) Pipelle suction (blind, random, smallvolume, single pass only); (b) curettage (nonblind, orientedfundus to cervix from bivalve hysterectomy specimen); and

(c) FFPE surgical hysterectomy specimens, as well as pairedendometrial hyperplasias and carcinomas and uninvolvedmyometrium from archived FFPE specimens. We also exam-ined DNA yield and DNA performance on the AffymetrixSNP 6.0 platform in a subset of samples.

2. Materials and Methods

2.1. Study Participants and Biological Samples. This study wasapproved by the University of Washington Human SubjectsDivision.

2.2. Fresh Tissue Collection. To assess whether quantity,quality, and performance of RNA and DNA obtained fromdifferent types of FF and FFPE endometrial samples fromthe same person varied by sample type, we enrolled eightwomen undergoing total abdominal hysterectomy for benignconditions in July and August of 2009. Indications forhysterectomy (not mutually exclusive) were irregular heavybleeding (𝑛 = 6); uterine prolapse (𝑛 = 1), and fibroids(𝑛 = 4). For each patient, three types of samples were colle-cted: (1) fresh Pipelle (CooperSurgical Inc.) biopsy; (2)fresh curettage biopsy; and (3) FFPE hysterectomy surgicalsections. Immediately following removal of the uterus, asingle pass with a Pipelle curette was performed and tissueswere placed into tissue culture media (minimum essentialmedia (MEM) with 5% fetal calf serum, 5mMHEPES buffer,and 10% DMSO) on ice for storage at −30∘C. No RNAlatersample was obtained as we had no data to guide whetherthe endometrial sample obtained with one pass would be ofsufficient volume for both diagnostic and research purposes;our primary focus was on the performance of the MEM-preserved specimens. The uterus was then bivalved and theendometrium was sharply curetted in two passes from thefundus to the junction of the uterus and the cervix. Onecurettage sample was placed into MEM, and the secondsample was placed into RNAlater (Qiagen, Valencia, CA,USA), a medium suitable for long-term storage of tissuesfor both RNA and DNA extractions. Both curettage sampleswere placed on ice at −30∘C until frozen for storage at −70∘C.The uterus was then sectioned and processed into FFPE hys-terectomy samples for histopathology. Each method of tissueharvesting potentially samples different cell types. Pipellebiopsies sample the endometrial lining, while the curettagebiopsies are comprised of predominantly endometrial tissuebut may contain myometrial cells. The FFPE hysterectomyspecimens contain both endometrial and myometrial celltypes.

We were able to obtain sufficient tissue for analysis from7 of 8 FF Pipelle (MEM), 8 of 8 FF curettage (MEM), 5 of 8 FFcurettage (RNAlater) samples, and all 8 hysterectomy samples(FFPE) (Table 1(a)).

2.3. Archival Tissue Collection. To assess quantity and qualityof DNA and RNA from older archived clinical samplesand performance on the WG-DASL and SNP 6.0 platforms,we selected 24 FFPE hysterectomy specimens: 6 complexhyperplasias; 6 atypical hyperplasias; and 12 carcinomas fromthe University of Washington Pathology Department. The

Obstetrics and Gynecology International 3

Table1:(a)B

enignhyste

rectom

yspecim

ens:RN

AandDNAyield

forF

Fpipelle

samples

preservedin

MEM

,FFcuretta

gesamples

preservedeither

inMEM

orin

RNAlater,and

FFPE

hyste

rectom

ysamples.(b)

ArchivedFF

PESpecim

ens:RN

AandDNAyieldandyear

ofcollectionforn

ormalandinvolved

tissues

amples,for

complex

hyperplasia

andatypicalhyperplasia

.(c)A

rchivedFF

PESpecim

ens:RN

AandDNAyieldandyear

ofcollectionfore

ndom

etria

lcarcino

mas

andno

rmalmyometriu

m.

(a)

Stud

yID

Specim

encollectiondate

Age

Menop

ause

status

Pipelle,M

EMCu

retta

ge,M

EMCu

retta

ge,

RNAlater

Hysterectom

y,FF

PE

DNA

yield (𝜇g)

Weighto

fRN

Atissue

(mg)

RNA

yield (𝜇g)

DNA

yield (𝜇g)

Weighto

fRN

Atissue

(mg)

RNA

yield (𝜇g)

DNA

yield (𝜇g)

Weighto

fRN

Atissue

(mg)

RNA

yield (𝜇g)

12009

46Pre

2431

6226.6

2650.1

3520

13.4

22009

50Peri

n/a∗

2923

8.5

2932.8

28.1

2116.8

32009

70Po

st101.9

n/a∗

n/a∗

23.4

n/a§

0†

X‡21

7.84

2009

45Pre

20.8

1925.4

40.3

3011.8

2124

18.2

52009

51Pre

37.7

3120.9

17.5

n/a§

0.5†

X‡25

2.7

62009

52Peri

6932

27.9

35.8

1825.1

X‡21

3.2

72009

47Peri

25.3

2744

48.7

2034

4921

4.4

82009

48Pre

46.3

2344

.617.2

2642.2

137.5

3726

Averagey

ield

46.4

27.4

35.4

27.3

24.8

32.7

54.1

23.8

11.6

∗

Not

enou

ghtissuer

emaining

forn

ucleicacid

extra

ction.

†

Not

enou

ghyield

tousetissue

ondo

wnstre

amapplications;<

1.2𝜇g.Th

esea

reno

tincludedin

thea

verage

yield.

‡

Notissuea

vailable.

§ Too

little

tissuetoweigh

accurately.

(b)

Com

plex

hyperplasia

,𝑛=6

Stud

yID

Specim

encollectiondate

Age

Menop

ause

status

Involved

Uninvolved

Weighto

fRNA

tissue(mg)

RNAyield

(𝜇g)

Weighto

fRNA

tissue(

mg)

RNAyield

(𝜇g)

92000

65Po

st16

12.6

1124.9

102001

53Po

st23

16.1

2818.3

112003

38Pre

2720.9

270.2†

122005

38Pre

361.6

3214

132008

55Po

st24

48.5

2519.1

142008

62Po

st25

3.1

175.7

Averagey

ield

25.2

17.1

23.3

16.4


(b)Con

tinued.

Atypicalhyperplasia

,𝑛=6

Stud

yID

Specim

encollectionsate

Age

Menop

ause

status

Involved

Uninvolved

Weighto

fRNA

tissue(mg)

RNAyield

(𝜇g)

Weighto

fRNA

tissue(

mg)

RNAyield

(𝜇g)

152001

32Pre

2110.8

2818.6

162003

54Pre

298.2

2718.2

172003

69Po

st24

17.8

2612.1

182005

43Peri

279.9

269.7

192005

54Pre/Peri

258

256.9

202008

72Po

st29

11.8

2621.4

Averagey

ield

25.8

11.1

26.3

14.5

(c) En

dometria

lCarcino

ma,𝑛=12

Stud

yID

Specim

enCollectionDate

Age

Menop

ause

status

Involved

Uninvolved

DNA

yield (𝜇g)

Weighto

fRN

Atissue

(mg)

RNA

yield (𝜇g)

DNA

yield (𝜇g)

Weighto

fRN

Atissue

(mg)

RNA

yield

(ng/ul)

211999

70Po

st14.7

1630

40.2

2343.1

221999

84Po

st89.7

3236.4

18.3

246.8

231999

80Po

st8.7

2620.9

36.8

2813.7

242004

77Po

st12.5

2529.6

51.3

296.1

252004

70Po

st33.2

3014.7

57.3

3411

262004

55Po

st83.04

2127.4

X‡X‡

X‡

272007

81Po

st40

.923

19.4

X‡X‡

X‡

282007

59Po

st17.38

3527.1

39.4

286.4

292007

63Po

st9.6

3111.1

X‡X‡

X‡

302009

56Peri

81.5

2432.1

35.2

2410.1

312009

44Pre

49.2

3218.1

X‡X‡

X‡

322009

52Po

st44

.632

39.7

X‡X‡

X‡

Averagey

ield

40.4

27.3

25.5

39.8

27.1

13.9

∗

Not

enou

ghtissuer

emaining

forn

ucleicacid

extra

ction.

†

Not

enou

ghyield

tousetissue

ondo

wnstre

amapplications;<

1.2𝜇g.Th

esea

reno

tincludedin

thea

verage

yield.

‡

Notissuea

vailable.

§ Too

little

tissuetoweigh

accurately.


samples were collected between 1999 and 2009 and stored foran average of 4.4 years (range 1 to 11 years) before tissue coreswere collected for RNA and DNA extraction.

All FFPE surgical histopathology slides were reviewed bya pathologist to identify and mark blocks with representativeand adequate areas of the various tissue types for RNAand DNA extraction. For the specimens with hyperplasia,uninvolved endometriumwasmarked for sampling 7–10mil-limeters away from the involved hyperplastic endometrium.For the specimens with carcinoma, uninvolved myometriumwas identified for sampling from separate blocks withoutcarcinoma. Samples were obtained from the paraffin blocksby coring the marked area(s) of the identified tissue with awarm 18-gauge needle. Approximately 15 cores were collectedper sample and the total core weight was recorded; up to37mg of paraffin-embedded tissue (range, 11–37mg). Corescollected from each tissue type were placed into separatemicrocentrifuge tubes, deparaffinized, and extracted usingtheQiagen Blood andTissueDNeasy spin columns accordingto the manufacturer’s protocol.

In total, we obtained FFPE tissue for 51 samples, includingthe benign hysterectomy samples described above (𝑛 = 8;Table 1(a)), as well as complex hyperplasia (𝑛 = 6 involved,𝑛 = 6 uninvolved); atypical hyperplasia (𝑛 = 6 involved, 𝑛 = 6uninvolved; Table 1(b)); and endometrial carcinoma (𝑛 = 12involved, 𝑛 = 7 uninvolved myometrium; Table 1(c)). Wewere unable to obtain uninvolvedmyometrial samples from 5of the 12 endometrial carcinoma samples because there wereno tissue blocks without carcinoma for these individuals.

2.4. RNA Extraction, Expression Array, and RT-PCR Vali-dation of Expression Array. RNA extraction was performedon samples with sufficient tissue. For the FF samples, thespecimens were split for RNA and DNA extraction, and forFFPE tissues, the weight of the cores for RNA extraction wasrecorded (Table 1(a)). Extractions were performed using theRecoverAll Total Nucleic Acid Isolation Kit (cat no. AM1975,Applied Biosystems/Ambion, Austin, TX, USA). RNA wasquantitated using the NanoDrop ND-1000 spectrophotome-ter. Quality was assessed bymeasuring the Ct threshold usingQuantiTect SYBR-Green RT-PCR mix using the protocolwith the WG-DASL kit (Illumina, San Diego, CA, USA) ona quantitative real-time PCR (qRT-PCR) system (AppliedBiosystems/Life Technologies 7900HT, Carlsbad, CA) inthe Fred Hutchinson Cancer Research Center (FHCRC)Genomics Core.

The amount of input RNA recommended by Illuminafor the WG-DASL (Illumina, San Diego, CA, USA) is 10–100 ng for RNA derived from FF samples and 50–200 ng forRNA derived from FFPE samples. Therefore, samples wereexcluded if they yielded less than 40 ng of RNA. Thirteen FFand 49 FFPE samples were assayed. To test reproducibilityof the assay, we randomly selected 8 FF samples (4 Pipelle;4 curettage) and 26 FFPE samples (4 benign; 3 complex, bothinvolved and uninvolved; 3 atypical involved and uninvolved;and 5 carcinoma, both involved and uninvolved) to run inreplicate for a total of 21 FF and 75 FFPE samples. Thesamples were processed and run on the WG-DASL arraysby the FHCRC Genomics Shared Resource according to the

manufacturer’s protocol. We examined whether the qualityand quantity of the RNA varied by tissue type and fixationmethod, and for the FFPE samples, whether RNA qualityvaried by year of sample processing. Various performancemetrics were examined, including the average number ofprobes detected (out of the 24,526 transcripts assayed), theaverage number of genes detected (out of the 18,391 genesassayed), the average signal p95, which is the 95th percentileof the probe intensities on the array, and the signal-to-noise ratio, which compares the strength of the signal tothe background signal. Using Genome Studio (Illumina, SanDiego, CA, USA), a detection 𝑃 value was calculated for eachtranscript, which represents the probability of observing agiven transcript if in fact the signal is not above the noise,with the background defined using negative control probes.We examined two levels of confidence:𝑃 < 0.01 and𝑃 < 0.05.While a larger number of probes are defined as detected for𝑃 < 0.05, they might not be as reproducible as the probe setdefined by 𝑃 < 0.01. Probe concordance is defined as thepercentage of the number of probes with matching detectedcalls (at 𝑃 < 0.01 or 𝑃 < 0.05) in two replicate samples,over the total number of probes detected in either of the twosamples.

To validate the relative expression values obtained byWG-DASL, expression levels of six genes were examinedby qRT-PCR in a subset of RNA samples (𝑛 = 44). Fivegenes (PTEN, CD79B, CD82, S100A4, and FOLR1) wereselected because they have been reported to be associatedwith endometrial hyperplasia or endometrial carcinoma(http://www.proteinatlas.org/) [6, 19], and one (KCNMA1)was selected because we observed a wide range of expres-sion levels (as assayed using WG-DASL) across the tissuesexamined. The 44 samples included 13 FF (7 MEM Pipelleand 6 MEM curettage) and 31 FFPE (8 benign hysterectomy,involved tissue from 6 complex hyperplasias, 5 atypicalhyperplasias, and 12 endometrial cancers). All 44 sampleswere run in duplicate, resulting in 88 samples for a total of 528data points. Averaged ratios of expression obtained by WG-DASL and qRT-PCR were compared for the six genes. Wedetermined the fold difference in expression of a given genebetween two sets of tissues by dividing the level of expressionin one tissue by the level in the other. We then comparedthe fold differences between the tissue types for measuresobtained by DASL versus those obtained by qRT-PCR. Thetwo different fold changes for a given gene were plotted as𝑋 and 𝑌 coordinates on a correlation graph and an 𝑅2 valuewas calculated to determine howwell the two sets ofmeasureswere correlated across all 6 genes.

2.5. DNA Extraction and Genotyping. For FF samples, DNAextractionwas performedon 20Pipelle and curettage samplesstored either in MEM (7 Pipelle and 8 curettage samples;one Pipelle did not have sufficient tissue for extraction) orRNAlater (5 of 8 curettage samples had sufficient tissue;no Pipelle samples were stored in RNAlater as the entirePipelle samples for researchwere placed inMEMtomaximizequantity). FF samples were minced with scalpel blades andsheared with NST buffer (146mM NaCl, 10mM Tris Base(pH 7.5), 1mM CaCl

2, 0.5mM MgSO

4, 0.05% Bovine serum


albumin (BSA), 21mM MgCl2, and 0.2% Igepal) with a 1 cc

syringe as needed prior to extraction using Puregene DNAIsolation Kit (Gentra Systems, Inc., Minneapolis, MN, USA).DNA was quantitated with PicoGreen (Quant-iT dsDNAAssay, Invitrogen, Carlsbad CA, USA).

For FFPE samples, we performed DNA extraction on 12endometrial cancers with paired uninvolved myometrium(𝑛 = 7; 5 cancers did not have corresponding blocks of uni-nvolved tissue). The samples were deparaffinized and extra-cted using the Blood and Tissue DNeasy spin columnsaccording to the manufacturer’s protocol (Qiagen, Valencia,CA, USA). DNA quality was assessed by measuring its sizeand concentration using a microfluidics-based platform offe-ring qualitative and semiquantitative analysis (Agilent 2100Bioanalyzer) with the DNA 7500 LabChip assay (AgilentTechnologies, Santa Clara, CA, USA) in the Fred HutchinsonCancer Research Center (FHCRC) Genomics Core.

To evaluate the performance of DNA derived from RNA-later and FFPE, we planned to genotype the DNA samplesfrom the 5 FF curettage samples preserved in RNAlaterand the 7 paired FFPE carcinomas and myometrium fromuninvolved tissue blocks on the Genome-Wide Human SNPArray 6.0 (Affymetrix, Santa Clara, CA, USA) according tothe standard Affymetrix protocol by HudsonAlpha Institutefor Biotechnology, Huntsville, AL, USA. According to thatprotocol, DNA was digested with Nsp I and Sty I restrictionenzymes and ligated to adaptors to amplify DNA prior togenotyping. However, the DNA from FFPE samples wasfragmented and failed to amplify. Therefore, these sampleswere not genotyped. DNA from the RNAlater samples wasresuspended at 50 ng/𝜇L prior to analysis on arrays.

3. Results

3.1. RNA Quantity and Quality and WG-DASL Array Perfor-mance and Reproducibility. We extracted RNA from 15 of 16FF MEM-preserved biopsy samples (7 Pipelle, 8 curettage)and 51 of 51 FFPE tissues. The extraction was consideredsuccessful if the sample yielded ≥1.2 𝜇g of RNA and had a260/280 ratio of 1.80 to 2.20. Extraction was successful for13/15 (87%) FF and 50/51 (98%) FFPE samples. All threefailures were attributed to low yields <1.2𝜇g (2 curettagesamples and one uninvolved FFPE sample from complexhyperplasia). One postmenopausal individual (patient IDno. 3) was found to have an atrophic endometrium. Notsurprisingly, there was insufficient tissue from both Pipelleand curettage (MEM and RNALater) for analysis (Table 1(a)).The yields from the FF tissues were higher than those fromthe FFPE tissues, but the extracted RNA was very pure for allof the sample types, as evidenced by the 260/280 ratios beingclose to 2.0, which is the 260/280 ratio for pureRNA (Table 2).

A prequalification qRT-PCR assay was performed to testthe quality of the RNA for optimal performance prior torunning theWG-DASL array. A Ct threshold above 29 cycleswas used as a cutoff to indicate poor quality samples. Whilethe FF samples performed slightly better as evidenced bylower numbers of Ct cycles than did the FFPE samples, all 63extracted samples passed the prequalification assay (Table 2).

The overall performance of the RNA samples on theWG-DASL array was excellent and was only slightly betterfor the FF samples than the FFPE samples with respect tothe 95th intensity percentile (p95) and signal-to-noise ratio.The average number of probes detected (𝑃 < 0.05) wassimilar for both FF and FFPE tissues, with 75% (18421/24526)detected for FF, 73% (17959/24526) detected for FFPE, and69% of genes detected for both groups (12780/18401 for FFand 12296/18401 for FFPE). The average number of probesdetected at 𝑃 < 0.05 and 𝑃 < 0.01, respectively, ranged from17,197 (70.1%) and 15,640 (63.8%) for benign hysterectomyFFPE tissue to 18,640 (76.0%) and 17,064 (69.6%) for FFPipelle tissue.The proportion of genes detectedwas lowest forboth benign hysterectomy FFPE tissue and FFPE carcinomaat 64.8%, and highest for Pipelle, at 70.2% (Table 2). ThePipelle and curettage sampleswere highly comparablewith anaverage expression level correlation 𝑅2 = 0.939. The averageexpression level correlations for FF Pipelle and FF curettagecompared to their matched FFPE hysterectomy samples weresimilar (0.690 and 0.692, resp.).

To assess the reproducibility of the WG-DASL array, weassessed the probe correlation for sample replicates. Expre-ssion levels were highly correlated for both the FF and FFPEreplicate samples (𝑅2 = 0.991 and 𝑅2 = 0.985, resp.). Probedetection concordance rates of the replicates were calculatedat 𝑃 values of <0.05 and <0.01, and all samples had highconcordance with an average of 96% gene overlap at both 𝑃values (Table 2).The lowest concordance rates were observedin the carcinomas at 94%. There was no difference in thequality of FFPE RNA samples, probe concordance acrossreplicates, RNA quality (260/280 ratio), or quantity (RNAyield), by year of sample collection (data not shown).

3.2. Validation of WG-DASL Array with qRT-PCR. Expres-sion of six genes was assayed using qRT-PCR on the sameRNA samples used for the WG-DASL arrays. Correlationsbetween fold differences in expression assayed using WG-DASL and qRT-PCR were evaluated for each gene betweentissue types (FF Pipelle and curettage, FFPE atypical andcomplex hyperplasia, FFPE cancer, and uninvolved myome-trium). Agreement between the two methods was good forPipelle and curettage (𝑅2 = 0.90) and cancer and myome-trium (𝑅2 = 0.82) but was poor for atypical and complexhyperplasia samples (𝑅2 = 0.02). To investigate whether theatypical or complex hyperplasia RNA samples had deterio-rated over time, we examined the fold change in expression ofthe panel of genes as determined byWG-DASL and qRT-PCRin the atypical and complex hyperplasia samples comparedto FFPE hysterectomy from the same women. The geneexpression changes between atypical hyperplasia and FFPEhysterectomy assayed using the WG-DASL and qRT-PCRwere well correlated (𝑅2 = 0.82), but this was not the case forcomplex hyperplasia compared to FFPE hysterectomy (𝑅2 =0.18). We did not find evidence of suboptimal qRT-PCR dueto technical issues (e.g., well position in the PCR block),suggesting the discrepancy was due to sample degradationspecific to the complex hyperplasia samples in the timebetween the WG-DASL and qRT-PCR analyses (36 weeks).


Table2:RN

Aqu

ality

metric

sand

assayperfo

rmance.

Benign

Com

plex

hyperplasia

Hyperplasiawith

atypia

Cancer

FFPipelle

FFCu

retta

geFF

PENormal

Involved

Normal

Involved

Normal

Involved

RNAEx

tractio

nNum

bersuccessfully

extracted

(overn

umbera

ttempted)∗

7/7

6/8

8/8

5/6

6/6

6/6

6/6

7/7

12/12

MeanRN

Ayield†

(𝜇g)

(range)

35.4

(20.9–

62.0)

32.7

(25.1–50.4)

11.6

(2.7–

26)

16.4

(5.7–

24.9)

17.1

(1.6–4

8.5)

14.5

(6.9–21.4

)11.1

(8.0–17.8

)13.9

(6.1–

43.1)

25.5

(11.1–39.7

)

Mean260/280ratio†

(range)

2.05

(2.01–2.08)

2.06

(2.05–2.09)

2.00

(1.9–2.0)

2.02

(1.93–2.05)

2.05

(1.91–2.1)

2.02

(1.97–2.05)

1.98

(1.8–2.03)

2.04

(1.93–2.09)

2.00

(1.9–2.1)

DASL

Pre-qu

alificatio

n‡

MeanCt

threshold(range)

15.7

(15.0–

16.4)

15.7

(15–18.7)

21.0

(18.0–

25.4)

20.9

(19.6

–22.5)

19.5

(18.2–20.7)

20.5

(19.3–21.3

)19.9

(20.0–

21.7)

21.5

(19–24.7)

21.0

(18.5–27.6)

DASL

Expressio

nArray

Array

passrat𝑒

§7/7

6/6

8/8

5/5

5/5

6/6

6/6

7/7

12/12

Averagen

umbero

fprobes

detected

(of24,526,𝑃<0.05)

18,640

(76.0%

)18,16

5(74.1%

)17,19

7(70.1%

)18,455

(75.2%

)18,244

(74.4%

)18,239

(74.4%

)18,446

(75.2%

)18,17

7(74.1%

)17,648

(72%

)

Averagen

umbero

fprobes

detected

(of24,526,𝑃<0.01)

17,064

(69.6

%)

16,637

(67.8

%)

15,640

(63.8%

)16,912

(69.0

%)

16,846

(68.7%

)16,647

(67.9

%)16,823

(68.6%

)16,442

(67.0

%)15,874

(64.7%

)

Averagen

umbero

fgenes

detected

(of18,40

1,𝑃<0.01)¶

12,917

(70.2%

)12,630

(68.6%

)11,931

(64.8%

)12,434

(67.6

%)

12,580

(68.4%

)12,642

(68.7%

)12,645

(68.7%

)12,348

(67.1%)

11,930

(64.8%

)

AverageS

ignalp

95¶

7,436

8,495

6,630

6,906

6,944

7,202

6,921

6,703

6,725

Sign

al-to

-noise

ratio

¶191

188

184

179

184

176

179

178

192

ReplicateC

oncordance

Num

bero

freplicates

runon

array

34

43

33

35

5

Correlationof

expressio

nlevels

(𝑟2

)0.99

0.99

0.99

0.99

0.99

0.98

0.99

0.99

0.98

Prob

econ

cordance,𝑃<0.05

0.97

0.98

0.96

0.96

0.96

0.95

0.96

0.96

0.94

Prob

econ

cordance,𝑃<0.01

0.98

0.98

0.97

0.96

0.97

0.96

0.96

0.96

0.94

∗

Extractio

nswered

efinedas

successfu

lifthe

yield

was>1.2𝜇gandthe2

60/280

ratio

was

between1.8

0and2.20,asp

ureR

NAhasa

260/280ratio

of2.0.

†

Exclu

ding

thethree

samples

thathadyields<1.2𝜇g.

‡

ACt

thresholdof

29or

belowqu

alifies

RNAforthe

DASL

array.

§ Allreplicates

also

passed.

¶ Inou

rhigh-throug

hput

facility,thea

verage

numbero

fgenes

detected

is10,000,the

averagep

95sig

nalis5

,000,and

thea

verage

signal-to-no

iseratio

is40

.


3.3. DNA Quantity, Quality, and Genotyping Success. DNAextraction was successfully performed on all 20 FF samplesand all 19 FFPE samples (12 carcinomas and 7 uninvolvedmyometrium)with excellent yields from all sample types.Theaverage DNA yield was considerably higher for MEM Pipelle(46.4 𝜇g) than MEM curettage (27.3𝜇g; Table 1(a)), thoughthe DNA yield from RNAlater curettage was the highest(54.1 𝜇g). The DNA yield from FFPE samples was morethan adequate for genomic studies (40.2𝜇g). Genotyping wasplanned only for the 5 RNAlater-stored curettage samplesand 14 of the FFPE samples (7 carcinomas and their paireduninvolved myometrium samples). However, because DNAfragmentation of the FFPE samples was detected, thesesamples were not genotyped. Of the five FF RNAlater-storedcurettage samples, one failed due to low input DNA. Theremaining four samples performed well, with call rates of98.2% to 99.3% (mean, 98.8%).

4. Discussion

This pilot study assessed the feasibility of using fresh andFFPE uninvolved, hyperplastic, and endometrial carcinomaspecimens collected during routine clinical care for high-throughput, array-based, and quantitative methodologies.We demonstrated the ability to extract high-quantity and-quality RNA from both FF and FFPE specimens for alltissue types and showed generally successful performanceon the WG-DASL platform. We also demonstrated thatDNA derived from fresh samples stored in RNAlater can beused successfully for genotyping on the Affymetrix SNP 6.0platform. The use of RNAlater has implications for simpli-fying clinical tissue collection, since samples can remain atroom temperature during surgery rather than stored on iceas is necessary for samples placed in MEM. The SNP 6.0array platform was not successful for DNA isolated fromFFPE carcinomas and myometrium, despite there being asubstantial volume of DNA. It is possible that our FFPE-derived DNA sample preparation protocol was not optimal;others have had success with the SNP 6.0 platform on DNAderived from FFPE tissue after implementing adjusted prepa-ration protocols to improve hybridization performance and amodified data analysis procedure [20]. The protocol we usedto collect DNA fromFFPE tissues yielded farmoreDNA thanis necessary for most genomic applications; future studiescould take a smaller number of cores for DNA extractionto preserve the tissue for additional assays. The techniquesvalidated in this pilot study have immediate potential to beused in samples from our existing longitudinal, retrospectivecohort [4, 8] and to ultimately be translated to future clini-cal applications with specimens collected during outpatientgynecologic visits for abnormal bleeding in women at risk forendometrial neoplasia.

Of particular interest is the assessment of the qualityand quantity of RNA extracted from Pipelle specimens, sincethis method of endometrial sampling is minimally invasiveand can be performed repeatedly over time. We demon-strated that single pass Pipelle tissue specimens (disorderedblinded specimens) are equivalent to curettage specimenswith regard to DNA and RNA yield and RNA performance

on the WG-DASL array. A limitation of our study was alack of direct comparison of WG-DASL results between FFPipelle and FFPE Pipelle samples. To our knowledge, RNAand DNA quality and quantity from Pipelle samples havenot been evaluated previously. Because we needed to assuresufficient endometrium for clinical diagnostic procedures,and the amount of tissue needed to extract high-qualityRNA and DNA was unknown, we conservatively choseto use the remaining sample for FF analysis. Given thatthe optimal Pipelle endometrial sampling process typicallyincludes multiple (rather than single) uterine passes, andthat we observed high RNA and DNA yields from the singlepass Pipelle specimens, we propose that tissue collected ineach pass could be split to use for diagnostic histopathologicevaluation and for research purposes.

We observed moderate success with RNA expressionassayed using the WG-DASL method. While some priorstudies have reported concordance between FF and FFPEsamples [13], others have described poor correlations at thegene or probe level [15, 16], noting that combinations ofgenes (particularly those already identified as members of apredictive signature, that is, for ovarian cancer subtypes basedon The Cancer Genome Atlas data) performed reasonablywell. Given that there are no gene expression signatures todate that identify endometrial hyperplasia subgroups that arelikely to respond to progestin, or are likely to progress, itis of critical importance to have complete and reliable geneexpression data.Thus, while theWG-DASL approach appearsto work reasonably well for validation, it is not as effectivefor discovery and is therefore not the ideal platform to use toidentify such subgroups.

Sequencing of both RNA and DNA derived from FFPEtissues is feasible, but performance of these methods willneed to be assessed within specific endometrial sample types[21, 22]. Development of novel analytic methods for qualitycontrol of RNA expression data generated from FFPE tissues,such as the quality control pipeline described by Waldronet al. [23], are particularly important as data from multiplestudy sites over longer time periods are combined. Addition-ally, as new methods are developed, such as a novel methodto simultaneously extract DNA, RNA, and microRNA froma single FFPE sample without splitting it [24], performanceof the methods will need to be evaluated for endometrialsamples.

5. Conclusions

Our study demonstrates that the Pipelle specimen, which is acommon preferred method of clinical endometrial samplingin women with abnormal bleeding, can be used effectivelyfor RNA expression studies and also provides a high yield ofDNA.We propose that in the future, while most of the Pipellewould be reserved for FFPE diagnostic assays, a portionof the sample could be retained for fresh tissue analysis.Until endometrial biomarker studies become a routine aspectof care, our study demonstrates that instituting researchpractices that collect fresh specimens as well as retaining themajority of the specimen for FFPE should not affect currentdiagnostic capabilities and health outcomes. RNA expression


data from FF Pipelle samples can be used to generate prog-nostic signatures, which can then be validated in archivedFFPE tissues. Interrogation of longitudinal endometrial FFPEtissue banks to better understand alterations that occur in theprogression of endometrial neoplasms is certainly feasible.

Acknowledgments

This study is supported by Grants NIH R01 HD 44813 andNIH P01 CA 91955 and the University ofWashington RoyaltyResearch Pilot Award.

References

[1] J. I. Sorosky, “Endometrial cancer,” Obstetrics and Gynecology,vol. 120, no. 2, part 1, pp. 383–397, 2012.

[2] L. S. Massad, M. H. Einstein, W. K. Huh et al., “2012 updatedconsensus guidelines for the management of abnormal cervicalcancer screening tests and cancer precursors,” Journal of LowerGenital Tract Disease, vol. 17, no. 5, supplement 1, pp. S1–S27,2013.

[3] R. J. Kurman, P. F. Kaminski, and H. J. Norris, “The behaviorof endometrial hyperplasia. A long-term study of “untreated”hyperplasia in 170 patients,” Cancer, vol. 56, no. 2, pp. 403–412,1985.

[4] S. D. Reed, K. M. Newton, R. L. Garcia et al., “Complexhyperplasia with and without atypia: clinical outcomes andimplications of progestin therapy,” Obstetrics and Gynecology,vol. 116, no. 2, part 1, pp. 365–373, 2010.

[5] J. V. Lacey Jr., O. B. Ioffe, B. M. Ronnett et al., “Endometrialcarcinoma risk among women diagnosed with endometrialhyperplasia: the 34-year experience in a large health plan,”British Journal of Cancer, vol. 98, no. 1, pp. 45–53, 2008.

[6] K. H. Allison, E. Tenpenny, S. D. Reed, E. M. Swisher, andR. L. Garica, “Immunohistochemical markers in endometrialhyperplasia: is there a panel with promise? A review,” AppliedImmunohistochemistry & Molecular Morphology, vol. 16, no. 4,pp. 329–343, 2008.

[7] K.H.Allison, S.D. Reed, L. F. Voigt, C.D. Jordan, K.M.Newton,and R. L. Garcia, “Diagnosing endometrial hyperplasia: whyis it so difficult to agree?” The American Journal of SurgicalPathology, vol. 32, no. 5, pp. 691–698, 2008.

[8] S. D. Reed, L. F. Voigt, K. M. Newton et al., “Progestin therapyof complex endometrial hyperplasia with and without atypia,”Obstetrics and Gynecology, vol. 113, no. 3, pp. 655–662, 2009.

[9] H. Liang, L. W. Cheung, J. Li et al., “Whole-exome sequencingcombined with functional genomics reveals novel candidatedriver cancer genes in endometrial cancer,” Genome Research,vol. 22, no. 11, pp. 2120–2129, 2012.

[10] C. Kandoth, N. Schultz, A. D. Cherniack et al., “Integratedgenomic characterization of endometrial carcinoma,” Nature,vol. 497, no. 7447, pp. 67–73, 2013.

[11] J. Y. Chung, T. Braunschweig, R. Williams et al., “Factorsin tissue handling and processing that impact RNA obtainedfrom formalin-fixed, paraffin-embedded tissue,” The Journal ofHistochemistry and Cytochemistry, vol. 56, no. 11, pp. 1033–1042,2008.

[12] C. April, B. Klotzle, T. Royce et al., “Whole-genome gene expre-ssion profiling of formalin-fixed, paraffin-embedded tissuesamples,” PLoS One, vol. 4, no. 12, Article ID e8162, 2009.

[13] L. Mittempergher, J. J. de Ronde, M. Nieuwland et al., “Geneexpression profiles from formalin fixed paraffin embeddedbreast cancer tissue are largely comparable to fresh frozenmatched tissue,” PLoS One, vol. 6, no. 2, Article ID e17163, 2011.

[14] N. Waddell, S. Cocciardi, J. Johnson et al., “Gene expressionprofiling of formalin-fixed, paraffin-embedded familial breasttumours using the whole genome-DASL assay,” Journal ofPathology, vol. 221, no. 4, pp. 452–461, 2010.

[15] M. G. Kibriya, F. Jasmine, S. Roy, R. M. Paul-Brutus, M. Argos,and H. Ahsan, “Analyses and interpretation of whole-genomegene expression from formalin-fixed paraffin-embedded tissue:an illustration with breast cancer tissues,” BMC Genomics, vol.11, no. 1, article 622, 2010.

[16] G. P. Sfakianos, E. S. Iversen, R. Whitaker et al., “Validation ofovarian cancer gene expression signatures for survival and sub-type in formalin fixed paraffin embedded tissues,” GynecologicOncology, vol. 129, no. 1, pp. 159–164, 2013.

[17] T. J. Clark, C. H. Mann, N. Shah, K. S. Khan, F. Song, andJ. K. Gupta, “Accuracy of outpatient endometrial biopsy inthe diagnosis of endometrial cancer: a systematic quantitativereview,” An International Journal of Obstetrics and Gynaecology,vol. 109, no. 3, pp. 313–321, 2002.

[18] F. P. Dijkhuizen, B. W. Mol, H. A. Brolmann, and A. P. Heintz,“The accuracy of endometrial sampling in the diagnosis ofpatients with endometrial carcinoma and hyperplasia: a meta-analysis,” Cancer, vol. 89, no. 8, pp. 1765–1772, 2000.

[19] J. E. Allard, J. I. Risinger, C. Morrison et al., “Overexpression offolate binding protein is associated with shortened progression-free survival in uterine adenocarcinomas,” Gynecologic Oncol-ogy, vol. 107, no. 1, pp. 52–57, 2007.

[20] M. Tuefferd, A. de Bondt, I. van denWyngaert, W. Talloen, andH. Gohlmann, “Microarray profiling of DNA extracted fromFFPE tissues using SNP 6.0 Affymetrix platform,” Methods inMolecular Biology, vol. 724, pp. 147–160, 2011.

[21] L. Waldron, P. Simpson, G. Parmigiani, and C. Huttenhower,“Report on emerging technologies for translational bioinfor-matics: a symposium on gene expression profiling for archivaltissues,” BMC Cancer, vol. 12, p. 124, 2012.

[22] H. Beltran, R. Yelensky, G. M. Frampton et al., “Targeted Next-generation Sequencing of Advanced Prostate Cancer IdentifiesPotentialTherapeutic Targets andDiseaseHeterogeneity,”Euro-pean Urology, vol. 63, no. 5, pp. 920–926, 2013.

[23] L. Waldron, S. Ogino, Y. Hoshida et al., “Expression profiling ofarchival tumors for long-term health studies,” Clinical CancerResearch, vol. 18, no. 22, pp. 6136–6146.

[24] A. Kotorashvili, A. Ramnauth, C. Liu et al., “Effective DNA/RNA co-extraction for analysis of microRNAs, mRNAs, andgenomic DNA from formalin-fixed paraffin-embedded speci-mens,” PLoS One, vol. 7, no. 4, Article ID e34683, 2012.

Submit your manuscripts athttp://www.hindawi.com

Stem CellsInternational

Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014


MEDIATORSINFLAMMATION

of


Behavioural Neurology

EndocrinologyInternational Journal of



Disease Markers


BioMed Research International

OncologyJournal of



Oxidative Medicine and Cellular Longevity


PPAR Research

The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014

Immunology ResearchHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014

Journal of

ObesityJournal of



Computational and Mathematical Methods in Medicine

OphthalmologyJournal of


Diabetes ResearchJournal of



Research and TreatmentAIDS


Gastroenterology Research and Practice


Parkinson’s Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttp://www.hindawi.com

Documents

Research Article Feasibility of RNA and DNA Extraction ...downloads.hindawi.com/journals/ogi/2013/576842.pdf · fresh or fresh frozen samples, as the purity and quantity of extracted